linux-kernel - Re: [PATCH] RISC-V: Remove IORESOURCE_BUSY flag for no-map reserved memory

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <f214bb0b-41d9-18cc-9f22-f3a7918689c9@linux.alibaba.com>
Date:   Wed, 11 May 2022 20:43:25 +0800
From:   Xianting Tian <xianting.tian@...ux.alibaba.com>
To:     David Hildenbrand <david@...hat.com>, paul.walmsley@...ive.com,
        palmer@...belt.com, aou@...s.berkeley.edu,
        akpm@...ux-foundation.org, anup@...infault.org,
        wangkefeng.wang@...wei.com, rppt@...nel.org, alex@...ti.fr,
        twd2.me@...il.com, seanjc@...gle.com, petr.pavlu@...e.com,
        atishp@...osinc.com
Cc:     linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org,
        guoren@...nel.org, jianghuaming.jhm@...baba-inc.com,
        Nick Kossifidis <mick@....forth.gr>
Subject: Re: [PATCH] RISC-V: Remove IORESOURCE_BUSY flag for no-map reserved
 memory


在 2022/5/11 下午8:37, Xianting Tian 写道:
>
> 在 2022/5/11 下午8:27, David Hildenbrand 写道:
>> On 11.05.22 13:24, Xianting Tian wrote:
>>> Commit 00ab027a3b82 ("RISC-V: Add kernel image sections to the 
>>> resource tree")
>>> added IORESOURCE_BUSY flag for no-map reserved memory, this casued
>>> devm_ioremap_resource() failed for the no-map reserved memory in 
>>> subsequent
>>> operations of related driver, so remove the IORESOURCE_BUSY flag.
>>>
>>> The code to reproduce the issue,
>>> dts:
>>>     mem0: memory@...00000 {
>>>                  reg = <0x0 0xa0000000 0 0x1000000>;
>>>                  no-map;
>>>          };
>>>
>>>     &test {
>>>         status = "okay";
>>>         memory-region = <&mem0>;
>>>     };
>>>
>>> code:
>>>     np = of_parse_phandle(pdev->dev.of_node, "memory-region", 0);
>>>     ret = of_address_to_resource(np, 0, &r);
>>>     base = devm_ioremap_resource(&pdev->dev, &r);
>>>     // base = -EBUSY
>>>
>>> Fixes: 00ab027a3b82 ("RISC-V: Add kernel image sections to the 
>>> resource tree")
>>> Reported-by: Huaming Jiang <jianghuaming.jhm@...baba-inc.com>
>>> Reviewed-by: Guo Ren <guoren@...nel.org>
>>> CC: Nick Kossifidis <mick@....forth.gr>
>>> Signed-off-by: Xianting Tian <xianting.tian@...ux.alibaba.com>
>>> ---
>>>   arch/riscv/kernel/setup.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/arch/riscv/kernel/setup.c b/arch/riscv/kernel/setup.c
>>> index 834eb652a7b9..71f2966b1474 100644
>>> --- a/arch/riscv/kernel/setup.c
>>> +++ b/arch/riscv/kernel/setup.c
>>> @@ -214,7 +214,7 @@ static void __init init_resources(void)
>>>             if (unlikely(memblock_is_nomap(region))) {
>>>               res->name = "Reserved";
>>> -            res->flags = IORESOURCE_MEM | IORESOURCE_BUSY;
>>> +            res->flags = IORESOURCE_MEM;
>>>           } else {
>>>               res->name = "System RAM";
>>>               res->flags = IORESOURCE_SYSTEM_RAM | IORESOURCE_BUSY;
>> I assume the "Reserved" part is essentially unused by the kernel 
>> correct?
>
> I think we may use it, actually we found the issue in our product 
> after merged kdump functionality.
>
> Actually, the code didn't add IORESOURCE_BUSY for no-map reserved 
> memory before 00ab027a3b82 merged,  so it is a typo for commit 
> 00ab027a3b82 to add IORESOURCE_BUSY?

This is arm64 code, which doesn't add IORESOURCE_BUSY for no-map 
reserved memory,

arch/arm64/kernel/setup.c

         for_each_mem_region(region) {
                 res = &standard_resources[i++];
                 if (memblock_is_nomap(region)) {
                         res->name  = "reserved";
                         res->flags = IORESOURCE_MEM;
                         res->start = 
__pfn_to_phys(memblock_region_reserved_base_pfn(region));
                         res->end = 
__pfn_to_phys(memblock_region_reserved_end_pfn(region)) - 1;

>
>>